Professional Documents
Culture Documents
www.jatit.org
E-mail: kartik_pandya@rediffmail.com
ABSTRACT
The objective of an Optimal Power Flow (OPF) algorithm is to find steady state operation point which
minimizes generation cost, loss etc. or maximizes social welfare, loadability etc. while maintaining an
acceptable system performance in terms of limits on generators’ real and reactive powers, line flow
limits, output of various compensating devices etc. Traditionally, classical optimization methods were
used to effectively solve OPF. But more recently due to incorporation of FACTS devices and
deregulation of a power sector, the traditional concepts and practices of power systems are superimposed
by an economic market management. So OPF have become complex. In recent years, Artificial
Intelligence (AI) methods have been emerged which can solve highly complex OPF problems. The
purpose of this paper is to present a comprehensive survey of various optimization methods used to solve
OPF problems.
Keywords: FACTS (Flexible A.C. Transmission system) devices, Optimal Power Flow (OPF), Power
System Deregulation.
450
Journal of Theoretical and Applied Information Technology
www.jatit.org
objective function and the constraints in each J.A.Momoh [12] presented an extension of basic
iteration and it is better than one which linearizes Kuhn-Tucker conditions and employing a
only objective function once. generalized Quadratic-Based model for OPF. The
F. Lima et al. [8] used Mixed Integer Linear conditions for feasibility, convergence and
Programming to conduct design study on the optimality are included in the construction of the
combinatorial optimal placement of Thyristor OPF algorithm. It is also capable of using
Controlled Phase Shifter Transformers (TCPSTs) hierarchical structures to include multiple
in large-scale power systems. It finds the number, objective functions and selectable constraints. The
network location and settings of phase shifters that generalized algorithm using sensitivity of
maximize system loadability under the DC load objective functions with optimal adjustments in the
flow model, subject to limits on the installation constraints yields a global optimal solution.
investment or total number of TCPSTs. Computational memory and execution time
Computation time is significantly lower than those required have been reduced.
of other published comparable cases. N. Grudinin [13] proposed a reactive power
optimization model that was based on Successive
Quadratic Programming (SQP) methods. Six
4. NEWTON-RAPHSON (NR) METHOD optimization methods were used to test the IEEE
30 bus and 278 bus systems. It is found that the
The necessary conditions of optimality referred developed SQP methods provide more fast and
to as the Kuhn-Tucker conditions are obtained in reliable optimization in comparison with the usual
this method. SLP method.
S.Chen et al. [9] proposed a new algorithm G.P.Granelli et al. [14] proposed Security-
based on Newton-Raphson (NR) method with constrained economic dispatch using dual
sensitivity factors incorporated to solve emission sequential quadratic programming. A dual feasible
dispatch in real-time. The Jacobian matrix and the starting point is achieved by relaxing transmission
B-coefficients have been developed in terms of the limits and then constraint violations are enforced
generalized generation shift distribution factor. So using the dual quadratic algorithm. It is compared
the penalty factor and the incremental losses are with SQP method of NAG routine. It presents
easily obtained. Execution time is lesser than that limited computation time and sufficiently good
of the conventional one. accuracy.
K.L.Lo et al. [10] proposed two Newton-like X.Lin et al. [15] integrated cost analysis and
load flow methods, the Fixed Newton method and voltage stability analysis using an OPF
the modification of the right-hand-side vector formulation for competitive market, which was
method for line outage simulation that is a part of solved using sequential quadratic programming.
contingency analysis. These two methods are Optimum reactive power dispatch are obtained
compared with the Newton-based full AC load under various voltage stability margin
flow method and the fast decoupled load flow to requirements in both normal and outage conditions
show their better convergence characteristics. for IEEE 14-bus test system.
X. Tong et al. [11] presented the semismooth A.Berizzi et al. [16] presented Security–
Newton-type algorithms for solving OPF Constrained Optimal Power Flow (SCOPF) to
problems. It treated general inequality constraints determine optimal setting and operation mode of
and bounded constraints separately. By UPFC and TCPAR. The solution of the OPF is
introducing a diagonal matrix and the nonlinear obtained by the use of the HP (Han-Powell)
complementarity function, the KKT system of the algorithm. It is an efficient method that solves
OPF is transformed equivalently to a system of nonlinear problems with nonlinear constraints
nonsmooth bounded constrained equations. The through the solution of successive quadratic
number of variables is less compared to other problems with linear constraints. It is applied to
method. The method saves computing cost. CIGRE 63-bus system and Italian EHV network.
A global solution is achieved, even adopting
different starting points.
5. QUADRATIC PROGRAMMING (QP)
METHOD
451
Journal of Theoretical and Applied Information Technology
www.jatit.org
452
Journal of Theoretical and Applied Information Technology
www.jatit.org
453
Journal of Theoretical and Applied Information Technology
www.jatit.org
power systems using genetic algorithm. The total transfer capability (TTC) of power
objective is to achieve the power system economic transactions between source and sink areas in
generation allocation and dispatch in deregulated deregulated power system. EP simultaneously
electricity market. The locations of the FACTS searches for FACTS locations, FACTS
devices, their types and ratings are optimized parameters, real power generations except slack
simultaneously. UPFC, TCSC, TCPST and SVC bus in source area, real power loads in sink area
are modeled and their investment costs are also and generation bus voltages.
considered. P.Attaviriyanupap et al. [38] presented a new
bidding strategy for a day-ahead energy and
reserve markets based on an EP. The optimal
8.4. MISCELLANEOUS AI METHODS bidding parameters for both markets are
determined by solving an optimization problem
Chowdhury et al. [34] proposed Expert System that takes unit commitment constraints such as
(ES) which was used in combination with a generating limits and unit minimum up/down time
transmission constrained economic dispatch to constraints into account. The proposed algorithm
provide real time security. The strategy for is developed from the view point of a generation
combining the ES with an economic dispatch company wishing to maximize a profit as a
which identifies the constraint violations in bus participant in the deregulated power and reserve
voltage magnitudes and in line flows, as well as markets. Separate power and reserve markets are
the set of optimal generations. The ES then considered, both are operated by clearing price
determines the best possible control measure using auction system.
rules on voltage and line flow control. The purpose T.Jayabarathi et al. [39] proposed the
of the expert system is to expeditiously remediate application of Classical Evolutionary Programming
voltage and branch over load problems. (EP), Fast EP and Improved FEP methods to solve
H.Mori et al. [35] presented a Parallel Tabu all kinds of economic dispatch problems such as
Search (PTS) based method for determining ED of generators with prohibited operating zones
optimal allocation of FACTS devices in (POZ), ED of generators with piecewise quadratic
competitive power systems. Available Transfer cost function (PQCF), combined economic-
Capability (ATC) was maximized with the FACTS environmental dispatch (CEED) and multi-area
devices. UPFC was modeled and concept of economic dispatch (MAED). The constraints
incremental load rate was used. The proposed considered are the power balance, generating
method was compared with Simulated Annealing, capacity, prohibited operating zones, area power
GA and Tabu Search methods. It is 1.95 and 2.68 balance, generation limits and tieline limits
times faster than TS and GA respectively. It is not constraints.
affected by the initial conditions and gave higher
quality solutions.
8.6. ANT COLONY OPTIMIZATION (ACO)
454
Journal of Theoretical and Applied Information Technology
www.jatit.org
mechanism is developed for the solution of decreased. The particle also adjusts its velocity
optimal unit commitment (UC) with probabilistic according to two extremes. One is the best position
spinning reserve determination. Total production of its own and the other is not always the best one
fuel costs, start-up costs of units in stage t, the of the group, but selected randomly from the
penalty cost imposed when any of constraints are group.
violated and the total accumulated cost from stage J.G.Vlachogiannis et al. [46] formulated the
0 to stage t. are included in objective function. The contributions of generators to the power flows in
security function approach is also applied to transmission lines as a multiobjective optimization
evaluate the desired level of system security. problem and calculated using a Parallel Vector
R.Meziane et al. [42] used ACO to solve the Evaluated Particle Swarm Optimization (VEPSO)
allocation problem involving the selection of algorithm. VEPSO accounts for nonlinear
electrical devices and the appropriate levels of characteristics of the generators and lines. The
redundancy to maximize system reliability of contributions of generators are modeled as
series-parallel topology, under performance and positions of agents in swarms. Generator
cost constraints. A universal moment generating constraints such as prohibited operating zones and
function (UMGF) approach is used by the ACO to line thermal limits are considered. It can obtain
determine the optimal electrical power network precise solutions compared to analytical methods.
topology. M.Saravanan et al. [47] proposed the
application of Particle Swarm Optimization to find
the optimal location, settings, type and number of
8.7. PARTICLE SWARM OPTIMIZATION FACTS devices to minimize their cost of
(PSO) installation and to improve system loadability for
single and multi-type FACTS devices. While
It is based on the ideas of social behavior of finding the optimal location, the thermal limit for
organisms such as animal flocking and fish the lines and voltage limit for the buses are taken
schooling. as constraints. TCSC, UPFC and SVC were
H.Yoshida et al. [43] proposed a Particle considered.
Swarm Optimization (PSO) for reactive power and
Voltage/VAR Control (VVC) considering voltage 9. SUMMARY OF THE PAPER
security assessment. It determines an on-line VVC The following table shows the suitable methods
strategy with continuous and discrete control for solving the various optimization problems of
variables such as AVR operating values of Electrical Engineering.
generators, tap positions of OLTC of transformers Objective
Suitable Reason to use
and the number of reactive power compensation function to be
method(s) that method
equipment. optimized
Jong-Bae Park et al. [44] suggested a Modified Economic Fast methods
Particle Swarm Optimization (MPSO) for LP, NR
dispatch
economic dispatch with nonsmooth cost functions. Economic Non-linear
A position adjustment strategy is proposed to dispatch with problem
provide the solutions satisfying the inequality AI
non-smooth
constraints. The equality constraint is resolved by cost function
reducing the degree of freedom by one at random. Economic – Suitable for
Dynamic search-space reduction strategy is Emission Fuzzy conflicting
devised to accelerate the process. The results dispatch objectives
obtained from the proposed method are compared Reactive Accurate
with those obtained by GA, TS, EP, MHNN, power NLP, QP,IP,AI methods
AHNN and NM methods. It has shown superiority optimization
to the conventional methods. Optimal Multiobjective
Cui-Ru Wang et al. [45] presented a Modified location of AI nonlinear
Particle Swarm Optimization (MPSO) algorithm to FACTS device problem
solve economic dispatch problem. In the new Social welfare QP, AI ---Do---
algorithm, particles not only studies from itself and Congestion ---Do----
the best one but also from other individuals. By AI
management
this enhanced study behavior, the opportunity to Security Stable
find the global optimum is increased and the NLP, IP
constrained convergence
influence of the initial position of the particles is
455
Journal of Theoretical and Applied Information Technology
www.jatit.org
456
Journal of Theoretical and Applied Information Technology
www.jatit.org
[10] K. L. Lo and Z. J. Meng, “Newton-like [22] Whei-Min Lin, S-J. Chen and Y.S Su, “An
method for line outage simulation”, IEE proc.- application of interior point based OPF for
Gener. Transm. Distrib., vol. 151, no. 2, pp. system expansion with FACTS devices in a
225-231, March 2004. deregulated environment”, 0-7803-6338-
[11] X. Tong and M. Lin, “Semismooth Newton- 8/00/$10.00 © 2000 IEEE, pp. 1407-1412.
type algorithms for solving optimal power [23] D.Xiaoying, W. Xifan, S. Yonghua and
flow problems”, in Proc. of IEEE/PES G.Jian, “The interior point branch and cut
Transmission and Distribution Conference, method for optimal power flow”, 0-7803-
Dalian, China, 2005, pp.1-7. 7459-2/02/$17.00 © 2002 IEEE, pp. 651-655.
[12] J. A. Momoh, “A generalized quadratic-based [24] Wei Yan, J. Yu, D. C. Yu and K. Bhattarai,
model for optimal power flow”, CH2809- “A new optimal reactive power flow model in
2/89/0000-0261 $1.00 © 1989 IEEE, pp. 261- rectangular form and its solution by predictor
267. corrector primal dual interior point method”,
[13] N. Grudinin, “Reactive power optimization IEEE Trans. Power Syst., vol. 21, no. 1, pp.
using successive quadratic programming 61-67, Feb. 2006.
method”, IEEE Trans. Power Syst., vol. 13, [25] Badrul H. Chowdhury, “Towards the concept
no. 4, pp. 1219-1225, Nov. 1998. of integrated security: optimal dispatch under
[14] G. P. Granelli and M. Montagna, “Security- static and dynamic security constraints”,
constrained economic dispatch using dual Electric Power Syst. Research, vol.25, pp.
quadratic programming”, Electric Power Syst. 213-225, 1992.
Research, vol 56, pp. 71-80, 2000. [26] N. Iwan Santoso and Owen T. Tan, “Neural-
[15] X. Lin, A. K. David and C. W. Yu, “Reactive net based real time control of capacitors
power optimization with voltage stability installed on distribution systems”, IEEE
consideration in power market systems”, IEE Trans. Power delivery, vol.5, no.1, pp. 266-
proc.-Gener. Transm. Distrib., vol. 150, no. 3, 272, Jan. 1990.
pp. 305-310, May 2003. [27] V. Miranda and J. T. Saraiva, “Fuzzy
[16] A. berizzi, M. Delfanti, P. Marannino, M. S. modeling of power system optimal load flow”,
Pasquadibisceglie and A. Silvestri, “Enhanced IEEE Trans. Power Syst., vol. 7, no. 2,
security-constrained OPF with FACTS pp.843-849, 1992.
devices”, IEEE Trans. Power Syst., vol. 20, [28] V. C. Ramesh and Xuan Li, “A fuzzy
no. 3, pp. 1597-1605, Aug. 2005. multiobjective approach to contingency
[17] J. A. Momoh and J. Zhu, “Multi-area power constrained OPF”, IEEE Trans. Power Syst.,
systems economic dispatch using nonlinear vol. 12, no. 3, pp.1348-1354, Aug. 1997.
convex network flow programming”, Electric [29] Narayana Prasad Padhy, “Congestion
Power Syst. Research, vol. 59, pp. 13-20, management under deregulated fuzzy
2001. environment”, IEEE International Conference
[18] D. Pudjianto, S. Ahmed and G. Strbac, on Electric Utility Deregulation,
“Allocation of VAR support using LP and Restructuring and Power Technologies
NLP based optimal power flows”, IEE proc.- (DRPT2004), Hong Kong, April 2004, pp.
Gener. Transm. Distrib., vol. 149, no. 4, pp. 133-139.
377-383, July 2002. [30] David C. Walters and Gerald B. Sheble,
[19] G. L. Torres and V. H. Quintana, “A jacobian “Genetic algorithm solution of economic
smoothing nonlinear complementarity method dispatch with valve point loading”, 92SM
for solving nonlinear optimal power flows”, in 414-3, IEEE/PES 1992 summer meeting.
Proc. Of 14th PSCC, Sevilla, Session 41, paper [31] P. H. Chen and H. C. Chang, “Large scale
1, June 2002, pp.1-7. economic dispatch by genetic algorithm”,
[20] A. K. Sharma, “Optimal number and location IEEE Trans. Power Syst., vol. 10, no. 4, pp.
of TCSC and loadability enhancement in 1919-1926, Nov. 1995.
deregulated electricity markets using [32] T. S. Chung and Y. Z. Li, “A hybrid GA
MINLP”, Int. J. of Emerging Electric Power approach for OPF with consideration of
Syst., vol. 5, issue. 1, pp. 1-13, 2006. FACTS devices”, IEEE Power Engineering
[21] S. Granville, “Optimal reactive dispatch Review, pp. 47-50, Feb. 2001.
through interior point methods”, IEEE Trans. [33] L. J. Cai, I. Erlich and G. Stamtsis, “Optimal
Power Syst., vol. 9, no. 1, pp. 136-146, Feb. choice and allocation of FACTS devices in
1994. deregulated electricity market using genetic
457
Journal of Theoretical and Applied Information Technology
www.jatit.org
458